Re: [net PATCH v2 6/6] eth: fbnic: Handle FW mailbox completions flagged with an error

On Fri, Sep 18, 2026 at 08:17:06AM -0700, Alexander Duyck wrote:

On Thu, Sep 17, 2026 at 2:12 PM [off-list ref] wrote:

...

The modification this way was intentional. Basically the most likely
place for us to ever encounter an issue with  DMA is on mailbox
initialization. Doing it this way reduces the driver init time as
instead of waiting 10s for the mailbox polling to fail we just burn
through the DMA failures and get to the result quickly indicating that
the mailbox is dead.

Doing the same for the other paths adds extra complexity as they are
interrupt driver versus the mailbox init which is synchronous. In
addition at mailbox init we can only have one completion in flight,
whereas for the other calls we can actually have multiple so without
context we won't know which completion it is we need to release.

Thanks for responding to Sashiko with your reasoning.

Reviewed-by: Simon Horman <horms@kernel.org>